Exercise 04: Observe model utilization in AI Foundry
Scenario
As you deploy AI models to production, it is important to monitor their performance and usage to ensure they are functioning as expected and to identify any potential issues. To assist with this, Microsoft Foundry provides built-in observability features that allow you to track model performance, usage metrics, and errors. In addition, Azure Monitor and Application Insights can be integrated with Foundry to provide deeper insights into model behavior and application performance. This will help Zava ensure that their multimodal AI shopping assistant is reliable and performs well for their customers.
In this exercise, you will learn how to use the observability features in Microsoft Foundry to monitor and trace the performance of your deployed AI models. You will explore the Foundry Application analytics dashboard, trace model requests and responses, and analyze telemetry data using Azure Application Insights.
Objectives
After you complete this exercise, you will be able to:
- Use the AI Foundry Application analytics dashboard to monitor the performance of your AI models in real-time
- Trace model requests and responses to diagnose issues
- Use Azure Application Insights to collect and analyze telemetry data
- Generate alerts based on model performance metrics
Duration
- Estimated Time: 60 minutes